Uncertainty quantification of spatially uncorrelated loads with a reduced-order stochastic isogeometric method

Chensen Ding, Kumar K. Tamma, Haojie Lian, Yanjun Ding, Timothy J. Dodwell, Stéphane P.A. Bordas

Research output: Contribution to journalArticlepeer-review

2 Scopus citations

Abstract

This work models spatially uncorrelated (independent) load uncertainty and develops a reduced-order Monte Carlo stochastic isogeometric method to quantify the effect of the load uncertainty on the structural response of thin shells and solid structures. The approach is tested on two demonstrative applications of uncertainty, namely, spatially uncorrelated loading, with (1) Scordelis–Lo Roof shell structure, and (2) a 3D wind turbine blade. This work has three novelties. Firstly, the research models spatially uncorrelated (independent) load uncertainties (including both their magnitude and/or direction) using stochastic analysis. Secondly, the paper advances a reduced-order Monte Carlo stochastic isogeometric method to quantify the spatially uncorrelated load uncertainty. It inherits the merits of isogeometric analysis, which enables the precise representation of geometry and alleviates shell shear locking, thereby reducing the model’s uncertainties. Moreover, the method retains the generality and accuracy of classical Monte Carlo simulation (MCS), with significant efficiency gains. The demonstrative results suggest that there is a cost, which is 3% of the time used by the standard MCS. Furthermore, a significant observation is made from the conducted numerical tests. It is noticed that the standard deviation of the output (i.e., displacement) is strongly influenced when the load uncertainty is spatially uncorrelated. Namely, the standard derivation (SD) of the output is roughly 10 times smaller than the SD for correlated load uncertainties. Nonetheless, the expected values remain consistent between the two cases.

Original languageEnglish (US)
JournalComputational Mechanics
DOIs
StatePublished - Mar 24 2021

Bibliographical note

Funding Information:
The authors would like to thank the Luxembourg National Research Fund for Intuitive modelling and SIMulation platform (IntuiSIM) (PoC17/12253887) and the National Science Foundation of China (81772025) for their financial support.

Publisher Copyright:
© 2021, Springer-Verlag GmbH Germany, part of Springer Nature.

Keywords

  • Monte Carlo simulation
  • Reduced-order models
  • Spatially uncorrelated load uncertainties
  • Stochastic isogeometric analysis
  • Uncertainty quantification

Fingerprint

Dive into the research topics of 'Uncertainty quantification of spatially uncorrelated loads with a reduced-order stochastic isogeometric method'. Together they form a unique fingerprint.

Cite this